r"""Record the value

Make models eval mode

use spectral normal layer to stable the training process

r"""Compute visualization images and save them to the disk. Args: path (str): Location of the file. data (dict): Data

save the current content to the HMTL file

Calculate losses, gradients, and update network weights; called in every training iteration

1x1 convolution

Flush, close possible files, and remove stdout/stderr mirroring.

Make models train mode

sh_coeff: [bz, 9, 3]

:param images: PIL list :return: numpy list

r"""Get rank of the thread.

warp vertices from world space to uv space vertices: [bz, V, 3] uv_vertices: [bz, 3, h, w]

Constructor

r"""Get the image save path. Args: subdir (str): Sub-directory under the main directory for saving the outputs.

bbox from detector and landmarks are different

-- rendering shape with detail normal map

Return: face_shape -- torch.tensor, size (B, N, 3) pts @ rot + trans Parameters: face_shape -- t

Insert a header to the HTML file Parameters: text (str) -- the header text

add images to the HTML file Parameters: ims (str list) -- a list of image paths txts (str list) -- a list of image

Computes the l1 loss between the ground truth keypoints and the predicted keypoints Inputs: kp_gt : N x K x 3 kp_pred: N x K x 2

Calculates the rotation matrices for a batch of rotation vectors Parameters ---------- rot_vecs: torch.tensor Nx3

3x3 convolution with padding

Function computes the distance between image patches in0 and in1 INPUTS in0, in1 - torch.Tensor object of shape Nx3xXxY - image p

:param images: PIL list :return:

Return: face_lms -- torch.tensor, size (B, 68, 2) Parameters: face_proj -- torch.tensor, size

Traverses the object name and returns the last (rightmost) python object.

r"""Get world size. How many GPUs are available in this job.

Draw 68 key points Args: image: the input image kpt: (68, 3).

image: 0-255, uint8, rgb, [h, w, 3] return: detected box list

r"""Save network weights, optimizer parameters, scheduler parameters to a checkpoint.

r"""Compute results images for a batch of input data and save the results in the specified folder. Args: data_loader (tor

r"""Move all tensors inside data to gpu. Args: data (dict, list, or tensor): Input data.

Calculates landmarks by barycentric interpolation Parameters ---------- vertices: torch.tensor BxVx3, dtype = torch.float32